The intonation of echo wh-questions in Ecuadorian Spanish
Clara Huttenlauch1, Sophie Egger1, Daniela Wochner1, Ingo Feldhausen2
1
2
Department of Linguistics, University of Konstanz, Germany
Institute for Romance Languages and Literature, Goethe University Frankfurt, Germany
{clara.huttenlauch|sophie.egger|daniela.wochner}@uni-konstanz.de,
[email protected]
Abstract
In Romance languages, neutral echo wh-questions and
counterexpectational (i.e., non-neutral or incredulous) echo
wh-questions are usually distinguished by tonal differences in
the nuclear contour. In this study, we show that Ecuadorian
Spanish does not differentiate between those question types in
terms of tonal targets, but by expanding the pitch range in the
nuclear region. Results drawn from a production experiment
based on semi-spontaneous speech with nine native speakers
show that neutral as well as non-neutral echo wh-questions are
realized with the same nuclear configurations, typically a
rising one (L* H%). However, counterexpectational whquestions are realized with a significantly larger pitch range
than their neutral counterparts (with a difference of 2.03
semitones). From a diatopic point of view, our data indicate
that there are prosodic differences in the realization of
questions between the Andean and the coastal region.
Speakers from the latter region realize rising as well as low
and falling nuclear configurations (L* H%, L* L% and H*
L%) and use a lower mean pitch. We hereby provide
intonational support for the well-known division between
tierras bajas ‘lowlands’ and tierras altas ‘highlands’ used for
capturing the dialectal phonological variation on the segmental
level of Latin-American Spanish.
Index Terms: echo wh-questions, incredulity, intonation,
pitch range, Ecuadorian Spanish, dialectal variation, Sp_ToBI
1. Introduction
Echo questions are subclasses of questions that fully or
partially repeat a prior utterance. Their main function is for the
addressee to clarify the content of the preceding utterance and
to elicit missing information when he or she failed to
understand what the speaker just expressed in the discourse [1]
(see [2] and [3] for additional functions). In a neutral echo whquestion (see (1) below), the hearer did not comprehend the
name Pedro properly and asks who came. However, the
question in (1) can also be interpreted as a non-neutral echo
question. In this case, the hearer did understand the
information properly, but is surprised about what has been said
since it contradicts his or her expectations (namely, that Pedro
was the one who came). Hence, he or she checks who came.
Echo wh-questions
Context:
Pedro vino.
Echo:
¿Quién vino?
(1)
(‘Pedro came’)
(‘Who came?’)
As (1) illustrates, the two types of echo wh-questions are
not distinguished syntactically, i.e., by alternations in word
order (see [2] and [4] for details on the possibility of using wh-
in-situ and the quotative particles que ‘that’ and si ‘whether’).
In many Spanish varieties the difference between these two
types of questions is marked in terms of intonation. In
Mexican Spanish, for example, neutral echo wh-questions
(NEQs) have a rising nuclear configuration (L* LH%), while
non-neutral or counterexpectational echo wh-questions (CQs)
are characterized by a rising-falling contour (L+H* L%; [5]).
In Venezuelan Spanish, NEQs have a falling nuclear contour
(H+L* L%) and CQs are realized by a rising contour (L+H*
H%; [6]). In Argentinian Spanish, NEQs have a falling
boundary tone, while CQs end at a mid-level (L+¡H* HL% vs.
L+¡H* M%; [7]). In Castilian Spanish, a (rising-)falling
contour is attested for NEQs ((L+)¡H* L%) and a rising
contour for CQs (L+¡H* HH%; [8]). The Castilian patterns
have already been described in the seminal work of EscandellVidal [2], [3] and [4], in which she offers detailed analyses of
the interrelation between syntax, semantics, pragmatics and
the intonation of echo-questions. Next to the tonal differences,
all varieties also distinguish NEQs from CQs by expanding the
pitch range of the CQs ([5], [6], [7] and [8]). The combination
of tonal movement and pitch excursion is also attested in other
Romance languages, such as Italian [9] or Catalan [10].
In Ecuadorian Spanish, however, no differences in the
tonal realization have been attested so far [11]. Both question
types are characterized by a rising-falling nuclear
configuration (L* HL%). While this strongly contrasts with
the pattern previously described for other Spanish and
Romance varieties, Ecuadorian Spanish also makes use of
pitch excursion [11]. CQs can be marked by an extended peak
value of the high tone within the nuclear configuration.
Despite the thorough study of [11], the overall concept of the
edited book in which the work appeared entailed that [11]
analyzed only a very limited dataset: Two subjects from Quito
uttered up to two echo wh-questions per type. For this reason,
we wonder whether the inexistence of tonal differences can be
replicated by means of a more broadly based study or whether
the findings are rather a result of the too restricted sample.
Two hypotheses (H1 and H2) guide our study:
• H1: Both NEQs and CQs are realized with a low nuclear
pitch accent (L*) and a rising-falling contour at the end
of the utterance (HL%; in accordance with [11]).
• H2: In contrast to NEQs, CQs may have a wider pitch
expansion within the nuclear configuration (again, in
accordance with [11]).
2. Experiment
To analyze the intonation patterns of CQs and NEQs in
Ecuadorian Spanish we conducted a production experiment
based on semi-spontaneous speech gathered by means of a
Discourse Completion Task [12] and [13]. In the prosodic
analysis, we concentrated on the nuclear configuration (i.e.,
the nuclear pitch accent and the sentence-final boundary tone),
the pitch range and the mean pitch throughout the utterance.
and no more than two contexts of the same condition occurred
in a sequence. Furthermore, we switched the overall order of
the contexts after six speakers to avoid possible position
effects.
2.1. Methods
2.1.3. Procedure
2.1.1. Participants
In the Discourse Completion Task, participants were asked to
spontaneously answer to a given situation, e.g., the contexts
illustrated in (2) and (3). The contexts were presented by the
first author for the recording sessions in Konstanz and by the
fourth author for the sessions in Frankfurt (see below). Each
session started with a short practice phase in order to
familiarize the participants with the procedure. Participants
were asked to produce their (semi-)spontaneous reactions as
natural as possible. Feedback was provided only in cases of
hesitations or complete misunderstandings. In such cases,
participants were asked to repeat the utterance.
We recorded nine monolingual native speakers of Ecuadorian
Spanish (four male) aged between 19 and 38 years (average
25.9 years, SD = 7.0). All of them participated voluntarily for
a small reimbursement and they were unaware of the purpose
of the study. Five of them originated from Quito, the capital of
Ecuador, and one from Tulcán. Both cities are in the Andean
region, henceforth, the highlands. The last three speakers grew
up at the coast in Guayaquil, henceforth, the lowlands. All
participants had been living/studying in Germany at the time
of the recordings (range between 4 months and 12 years) and
are learning German as a foreign language.
2.1.2. Materials
In order to compare the nuclear configuration and the pitch
range of CQs and NEQs we generated 12 pairs of short
contexts, one evoking the production of a non-neutral
question, the other one favoring a neutral reading (see (2) and
(3) below). Contexts that elicit CQs were constructed in
parallelism to the one for counterexpectational echo whquestions used in the questionnaire of the Atlas of Spanish
Intonation [14]. Comparable segmental environments are
desired to allow for a thorough prosodic analysis. Thus, we
created contexts to evoke neutral wh-questions with similar
words and parallel syntax as in the non-neutral condition.
Possible target utterances elicited by contexts (2) and (3) are
shown in brackets. The contexts were controlled for eliciting
mostly sonorant material within the region of the nuclear
configuration (such as (se) llama ‘is called’ in the given
examples) in order to allow for a better pitch tracking.
CQ:
Te comentan que la madre de un conocido se
llama Zoila Vaca. No puedes creer que alguien se
llame así y preguntas muy incrédulo cómo se
llama. (Possible answer: ¿Cómo se llama?)
‘You are told that the name of a friend’s mother is
Ima Cow (pronounced: I’m a cow). You can’t
believe that this is someone’s name and you ask
incredulously what her name is.’ (Possible answer:
What’s her name?)
(2)
NEQ:
En una fiesta te presentan a la tía de tu
novio/novia pero no sabes si has entendido bien el
nombre. Pregúntale cómo se llama (Possible
answer: ¿Cómo se llama?)
‘At a party, you are introduced to your partner’s
aunt but you are not sure whether you understood
her name properly. Ask for her name.’ (Possible
answer: What’s her name?)
(3)
The sessions of seven of the participants took place in a
sound-attenuated cabin at the PhonLab of the University of
Konstanz. Productions were recorded using an MXL 990
condenser microphone and a Tascam HD P2 portable stereo
audio recorder. In addition, two participants were recorded in a
sound-attenuated booth at the phonetics laboratory at the
Goethe University Frankfurt using an Edirol by Roland, R-44
solid-state four-channel portable field recorder. A sampling
rate of 44.1 kHz and a resolution of 16 Bit was used in all
recordings.
2.1.4. Data treatment & analysis
Overall, we recorded 248 CQs and NEQs. 59 items had to be
excluded due to glottalization or infelicitous answers (e.g., the
context was not understood properly even after repetition).
The remaining 189 productions (CQ: 91; NEQ: 98) were
analyzed phonologically and phonetically using Praat [15].
The files were labeled segmentally at the level of
morphological word and syllable. Pitch accents and boundary
tones were labeled following the guidelines of Sp_ToBI
presented in [16]. Pitch values and the calculated pitch range
in semitones (st) were extracted automatically using a Praat
script. Pitch was extracted within the range of the nuclear
configuration at manually set minimum and maximum points.
2.2. Results
2.2.1. Intonation contours
Additionally, we included 12 filler trials with neutral
contexts to elicit simple declarative constructions.
Furthermore, six practice items were constructed. They
consisted of similar contexts, evoking CQs, NEQs and
declaratives. The order of the contexts was pseudorandomized
The results across all speakers and regions show that both CQs
and NEQs are predominantly realized with a rising nuclear
configuration consisting of a low nuclear accent followed by a
high boundary tone (L* H%; CQ: 57%, N=52; NEQ: 46%,
N=45). For examples of rising contours, see Figure 1, left and
mid panel. The second most common contour across both
question types displays a low nuclear accent followed by a
downstepped high boundary tone (L* !H%; CQ: 6%, N=5;
NEQ: 17%, N=17). The third and fourth most common
nuclear configurations are L* L% (CQ: 9%, N=8; NEQ: 7%,
N=7) and H* L% (CQ: 8%, N=7; NEQ: 6%, N=6)
respectively. Moreover, we find additional tonal
configurations. The percentage distribution of the attested
nuclear contours is shown in Figure 2. With respect to the
attested boundary tones, our data show a predominant use of
H% (CQ: 65%, N=59; NEQ: 50%, N=49). The second most
common boundary tone is !H% (CQ: 14%, N=13; NEQ: 27%,
L+H*
ke
¿Qué
L+H*
pi
ðjo
pidió
L*
mi
300
250
200
150
100
50
Frequency (Hz)
Frequency (Hz)
Frequency (Hz)
300
250
200
150
100
50
H%
H*
ɣel
ke
Miguel?
0
L*
ko
0.9705
H%
L+<H*
mjo
¿Qué
kwan
comió?
0
L*
ðo
ʝe
¿Cuándo
0.4615
Time (s)
300
250
200
150
100
50
L%
ɣo
llegó?
0
0.7124
Time (s)
Time (s)
Figure 1: Waveforms, spectrograms and smoothed F0-contours. Left: CQ ¿Qué pidió Miguel? ‘What ordered Miguel?’
(rising nuclear contour, male speaker, highlands). Mid: NEQ ¿Qué comió? ‘What did s/he eat?’ (rising nuclear contour,
same speaker as before). Right: CQ ¿Cuándo llegó? ‘When did s/he arrive?’ (low nuclear contour, male speaker, lowlands).
N=26) followed in number of productions by L% (CQ: 19%,
N=17; NEQ: 17%, N=17). The falling boundary tone HL%
occurs in CQs only in 2% (N=2) and in NEQs in 6% (N=6).
Distribution in %
60%
CQ
50%
NEQ
40%
30%
20%
10%
2.2.2. Pitch range and mean pitch
Re
st
L*
H
L* %
!H
%
L*
L%
H
*
L+ L%
H
*
!H
L* %
H
H L%
+L
*
L+ H%
H
*
H
H %
*
!H
%
0%
Nuclear contour
Figure 2: Distribution of the nuclear configurations in
CQs and NEQs across all speakers and regions.
80%
Distribution in %
CQ
NEQ
60%
40%
20%
0%
While L* H% is by far the most common configuration in the
productions by the highland speakers (CQ: 69%, N=43; NEQ:
54%, N=35), it is less so for the lowland speakers (CQ: 31%,
N=9; NEQ: 30%, N=10). Next to L* H%, speakers from the
lowlands also realize L* L% and H* L% with a comparatively
high frequency (L* L%; CQ: 21%, N=6; NEQ: 12%, N=4 [see
Figure 1, right panel] and H* L%; CQ: 17%, N=5; NEQ: 15%,
N=5). By comparison, speakers from the highlands produce
L* L% in 3% (N=2) of the CQs and in 5% (N=3) of the NEQs
and H* L% in 3% (N=2) of the CQs and in 2% (N=1) of the
NEQs (see Figure 3).
L* H%
L* L%
H* L% L* H%
Highlands
L* L%
H* L%
Lowlands
Nuclear contour
Figure 3: Distribution of rising (L* H%), low (L* L%)
and falling (H* L%) nuclear contours in CQs and
NEQs in highlands and lowlands.
Beyond that, the data reveal diatopic differences in the
realization of the nuclear configurations. Overall we analyzed
127 utterances from highland speakers (CQ: 62; NEQ: 65) and
62 utterances from lowland speakers (CQ: 29; NEQ: 33).
No clear tonal differences between the intonation contours of
the two question types investigated are found. However, an
additional analysis of the pitch range (measured in semitones,
st) within the nuclear configuration and the mean pitch
throughout the questions reveals acoustic differences between
CQs and NEQs.
Pitch range measured in the nuclear configuration was
statistically analyzed using a linear mixed effects regression
model (R-version 3.1.2) with question type (CQ vs. NEQ) as
fixed factor and participants and items as crossed random
factors, allowing for random adjustments of intercepts and
slopes [17]. To ensure the validity of the model, data points
with residuals beyond 2.5 standard deviations were removed
and the model was refitted. P-values were calculated using the
Satterthwaite approximation in the R-package lmerTest.
Results for pitch range of the nuclear configuration show a
significant main effect of question type. Participants realize
CQs with larger pitch ranges than NEQs (ß=2.03, SE=0.39,
t=5.19, p<0.00001). The difference in the nuclear
configuration of CQs across all speakers is on average 2.03 st
larger than in NEQs (CQ: 8.96 st vs. NEQ: 6.93 st), as
visualized in Figure 4 for L* H% contours. Regarding the
geographical difference (highlands vs. lowlands), both speaker
groups show this difference in pitch range. Speakers from the
highlands produce the CQs with a 1.98 st wider pitch range
than the NEQs (ß=1.98, SE=0.42, t=4.68, p<0.00001, CQ:
9.15 st vs. NEQ: 7.17 st). Speakers from the coastal region
show a 2.09 st larger pitch range in NEQs (ß=2.09, SE=0.80,
t=2.61, p<0.05) than in CQs (CQ: 8.55 st vs. NEQ: 6.46 st).
Differences in the mean pitch throughout the utterances
were also statistically analyzed using a linear mixed effects
regression model (R-version 3.1.2) with the same specification
and procedure as above (with question type as fixed factor and
participants and items as crossed random factors). Again, data
points with residuals beyond 2.5 standard deviations were
removed and p-values were calculated using the Satterthwaite
approximation in the R-package lmerTest.
350
Nuclear configuration
L* H%
Average f0 (Hz)
300
250
200
150
10
15
20
25
Normalized time
Question type: CQ NEQ
30
Figure 4: Time normalized F0-contour averaged
across male and female speakers and both
regions of the most frequent nuclear
configuration (L* H%) in CQs and NEQs.
Results for mean pitch show a significant main effect of
question type, as well. Productions of CQs show a higher F0mean than those of NEQs (ß=14.95, SE=3.13, t=4.78,
p<0.00001). The mean pitch in CQs is on average 15 Hz
higher than in NEQs (CQ: 210 Hz vs. NEQ: 195 Hz).
However, in a further analysis, we ran separate models for
participants from the highlands and the lowlands, to account
for possible diatopic effects and found that the main effect of
question type on the produced mean F0 within the utterances
only holds true for participants who originated from the
highlands (ß=21.13, SE=3.27, t=6.46, p<0.00001). This
speaker group produces CQs with an average mean pitch of
264 Hz, whereas the NEQs show a mean pitch of 243 Hz.
There is no such effect of question type in the productions of
participants from the lowlands (ß=2.81, SE=6.34, t=0.44,
p=0.6). The difference in the mean pitch of the productions of
lowland speakers is only 3 Hz (CQ: 149 Hz vs. NEQ: 146 Hz).
3. Discussion
The results of the production experiment reveal that H1 can
only partly be verified. In accordance with [11] we find the
same nuclear contour in the productions of both NEQs and
CQs. However, our results do not show a rising-falling
boundary tone (HL%), as claimed by O’Rourke [11], but a
rising boundary tone with a preceding low pitch accent (L*
H%). Additionally, 17% of the neutral questions end in a
shallower version of this contour, namely in a high plateau (L*
!H%), which may be due to the narrower pitch range
expansion across the nuclear contour. This suggests that
incredulity is not expressed by means of tonal differences in
Ecuadorian Spanish.
Interestingly, we find a difference in the tonal realization
of the nuclear configuration between speakers from the
highlands and the lowlands. The group from the highlands
prefers L* H%, while the lowland group also frequently
produces L* L% and H* L% next to L* H%. This result can
be interpreted as possible intonational support for the well-
known division between tierras bajas (‘lowlands’) and tierras
altas (‘highlands’) - terms that are used to capture the dialectal
phonological variation on the segmental level of Latin
American Spanish [18].
Regarding H2, our results support the findings of [11] that
a difference between CQs and NEQs is established by a wider
pitch expansion within the nuclear contour of CQs. Our
analysis reveals that the pitch range difference between CQs
and NEQs is significant for speakers of both geographical
regions. This is in line with findings from other Romance
languages that equally mark non-neutral questions by a wider
pitch range ([9], [10], [19] and [20]). As a basic principle, the
intonational meaning of CQs is conveyed acoustically, through
differences in the pitch range of the nuclear configuration.
As for the mean pitch, our data show that only speakers
from the Andean highlands produce CQs with a significantly
higher mean pitch than NEQs. Speakers from the coastal
region do not use mean pitch to differentiate between CQs and
NEQs. This can be interpreted as a further diatopic difference
between the highlands and the lowlands of Ecuador (along
with the differences in the tonal realization of the nuclear
configuration). However, these findings also might be due to a
gender effect: five out of the six participants from the
highlands were female and all speakers from the lowlands
were male. Therefore, we cannot exclude a possible effect of
gender rather than region. Nevertheless, differences in mean
pitch seem to be a possible factor for speakers to go back to in
order to discriminate between CQs and NEQs, signaling
incredulity by raising the pitch level.
4. Summary and Outlook
In this study, we investigated whether neutral echo whquestions (NEQs) and counterexpectational echo wh-questions
(CQs) in Ecuadorian Spanish can be distinguished by
intonational means. Our results suggest that speakers do not
differentiate between CQs and NEQs by means of tonal
movement. Both question types were mainly realized with a
low-rising contour (L* H%), with some diatopic differences
that we interpret as an intonational reflection of the classical
diatopic division of Latin American Spanish in highlands and
lowlands. The finding of a low-rising contour, which is also
attested for CQs in other Romance varieties (e.g. in [6] and
[8]), contradicts observations made by [11], who found that a
rising-falling contour is used to mark those types of questions.
While no tonal differences can be attested in the data, our
results show a clear distinction regarding the pitch range in the
nuclear contour. This suggests that speakers of Ecuadorian
Spanish (highlands and lowlands) express the difference
between CQs and NEQs by means of pitch range expansion.
Additionally, speakers may use a higher mean pitch
throughout the utterance in order to express incredulity.
However, a significant difference in the mean pitch is
confirmed only for the Andean region by our data. For future
work, it will be conducive to scrutinize a data set that carefully
controls for sex and origin of the participants. This allows to
carefully examine whether mean pitch differences in the
realization of CQs and NEQs result from gender effects or
differences in the geographical region. In addition, a higher
number of participants may be considered. Furthermore, a
perception experiment will help to show whether the
differences in pitch range and mean pitch are a salient factor
for hearers to discriminate between CQs and NEQs.
5. References
[1]
E.-J. Noh, "Echo questions: Metarepresentation and pragmatic
enrichment," Linguistics and Philosophy, vol. 21, pp. 603-628,
1998.
[2] V. Escandell-Vidal, "Intonation and procedural encoding: The
case of Spanish interrogatives," in Current Issues in Relevance
Theory, V. Rouchota and A. H. Jucker, Eds., Amsterdam: John
Benjamins, 1998, pp. 169-203.
[3] V. Escandell-Vidal, "Echo-syntax and metarepresentations,"
Lingua, vol. 112, pp. 871-900, 2002.
[4] V. Escandell-Vidal, "Los enunciados interrogativos: Aspectos
semánticos y pragmáticos," in Gramática Descriptiva de la
Lengua Española. vol. 3, I. Bosque and V. Demonte, Eds.,
Madrid: Espasa Calpe, 1999, pp. 3929-3991.
[5] C. de-la-Mota, P. M. Butragueño, and P. Prieto, "Mexican
Spanish intonation," in Transcription of Intonation of the
Spanish Language, P. Prieto and P. Roseano, Eds., München:
Lincom Europa, 2010, pp. 319-350.
[6] L. Astruc, E. Mora, and S. Rew, "Venezuelan Andean Spanish
intonation," in Transcription of Intonation of the Spanish
Language, P. Prieto and P. Roseano, Eds., München: Lincom
Europa, 2010, pp. 191-226.
[7] C. Gabriel, I. Feldhausen, A. Pešková, L. Colantoni, S.-A. Lee,
V. Arana, and L. Labastía, "Argentinian Spanish intonation," in
Transcription of Intonation of the Spanish Language, P. Prieto
and P. Roseano, Eds., München: Lincom Europa, 2010, pp. 285318.
[8] E. Estebas-Vilaplana and P. Prieto, "Castilian Spanish
intonation," in Transcription of Intonation of the Spanish
Language, P. Prieto and P. Roseano, Eds., München: Lincom
Europa, 2010, pp. 17-48.
[9] B. Gili Fivela, C. Avesani, M. Barone, G. Bocci, C. Crocco, M.
D'Imperio, R. Giordano, G. Marotta, M. Savino, and P.
Sorianello, "Intonational phonology of the regional varieties of
Italian," in Intonation in Romance, S. Frota and P. Prieto, Eds.,
Oxford: Oxford University Press, 2015, pp. 140-197.
[10] P. Prieto, J. Borràs-Comes, T. Cabré, V. Crespo-Sendra, I.
Mascaró, P. Roseano, R. Sichel-Bazin, and M. Vanrell Bosch,
"Intonational phonology of Catalan and its dialectal varieties," in
Intonation in Romance, S. Frota and P. Prieto, Eds., Oxford:
Oxford University Press, 2015, pp. 9-62.
[11] E. O’Rourke, "Ecuadorian Andean Spanish intonation," in
Transcription of Intonation of the Spanish Language, P. Prieto
and P. Roseano, Eds., München: Lincom Europa, 2010, pp. 227253.
[12] S. Blum-Kulka, J. House, and G. Kasper, "Investigating crosscultural pragmatics: An introductory overview," in CrossCultural Pragmatics: Requests and Apologies, S. Blum-Kulka, J.
House, and G. Kasper, Eds., Norwood: Ablex, 1989, pp. 1-34.
[13] P. Prieto and P. Roseano, Eds., Transcription of Intonation of the
Spanish Language. München: Lincom Europa, 2010.
[14] P. Prieto and P. Roseano. (2009-2010). Atlas interactivo de la
entonación
del
español.
Available:
http://prosodia.upf.edu/atlasentonacion/
[15] P. Boersma and D. Weenink, "Praat: doing phonetics by
computer," 5.3.14, 1992-2015.
[16] J. I. Hualde and P. Prieto, "Intonational variation in Spanish:
European and American varieties," in Intonational Variation in
Romance, S. Frota and P. Prieto, Eds., Oxford: Oxford
University Press, 2015, pp. 350-391.
[17] D. J. Barr, R. Levy, C. Scheepers, and H. J. Tily, "Random
effects structure for confirmatory hypothesis testing: Keep it
maximal," Journal of Memory and Language, vol. 68, pp. 255278, 2013.
[18] J. M. Lipski, Latin American Spanish. London: Longman, 1994.
[19] S. A. Lee, F. Martínez-Gil, and M. E. Beckman, "The
intonational expression of incredulity in absolute interrogatives
in Buenos Aires Spanish," in Selected Proceedings of the 4th
Conference on Laboratory Approaches to Spanish Phonology,
M. Ortega-Llebaria, Ed., Somerville, MA: Cascadilla, 2008, pp.
47-56.
[20] V. Crespo-Sendra, M. d. M. Vanrell, and P. Prieto, "Informationseeking questions and incredulity questions: Gradient or
categorical contrast," in Proceedings of the 5th International
Conference on Speech Prosody, May 11-14, Chicago, USA,
Proceedings, 2010, pp. 11-14.